Special Issue on "Fast Energy-Minimization-Based Imaging and Vision Techniques"
EXTENDED DEADLINE: 31 March 1999
Call for Papers
Energy-minimization methods are powerful tools in all domains of imaging
and computer vision. Many of them descend from Bayesian or variational
approaches to solve the related inverse problems.
Many numerical algorithms implementing these methods have been developed in
recent years for several applications, but, because of their exceedingly
high
computational complexity, their practical interest has been limited to
those cases where real-time performance is not required. However, there are
many applications for which the high quality of the solutions achievable
with these methods is strongly desirable, and true real-time is not a
strict constraint. Indeed, the term 'real-time' often has a relative
meaning, depending on the application considered, the actual requirement
being to have a 'reasonable' elapsed time. These considerations
notwithstanding, the speed performance of many energy-minimization
algorithms is at present not sufficient for most applications. On the other
hand, the development of computing power both in dedicated and
general-purpose hardware is about to enable us to take some of these
techniques to practical usefulness. This justifies, from a practical point
of view, the continued research interest in energy-minimization methods.
Two main strategies can be identified in order to face these problems. From
an architectural point of view, an effort should be made to fully exploit
existing architectures for the implementation of the algorithms, or to
design special hardware best suited for particular tasks. From an
algorithmic point of view, the search for new mathematical models and/or
computational schemes should be directed towards a better tractability of
the problems. Moreover, the generality of some approaches can be reduced to
obtain algorithms that are either intrinsically less expensive or more
suitable for particular high-performance machines.
All the contributions to solve (or approaching a solution of) one of the
problems raised above are welcome to this special issue. A (not exhaustive)
list of suggested sub-topics is the following:
1. Algorithmic aspects
1.1 Mathematical models
1.2 Fast numerical procedures
1.3 Specialized algorithms
1.4 Parallel implementations
1.5 Optimization of cooperating parallel and serial processes
2. Architectural aspects
2.1 Hardware development methodologies
2.2 Distributed computing approaches
2.3 Innovative architectures
2.4 VLSI implementations
3. Applications
3.1 Pattern recognition
3.2 Image segmentation
3.3 Autonomous vehicle guidance
3.4 Robot motion control
3.5 Remote sensing
3.6 Medical imaging
3.7 Industrial inspection
3.8 Visual data bases
3.9 Image coding
All enquiries can be addressed to the guest editor, Emanuele Salerno, at
the following email address: e.salerno@iei.pi.cnr.it
Manuscript Submission
Authors should send five copies of their manuscripts to Dr. E. Salerno, at
the address below. Each manuscript should contain a cover page with the
title and an abstract, and the indication of the two sub-topics that best
match the subject treated in the paper (as said, the above list is not
exhaustive).
Emanuele Salerno (Real-Time Imaging)
IEI-CNR
Via Santa Maria, 46
I-56126 Pisa, Italy
Submission deadline
All contributions should be received by 31 March 1999.
________________________________________________________
Dr. Anna Tonazzini
Istituto di Elaborazione della Informazione
Consiglio Nazionale delle Ricerche
Via S. Maria, 46
I-56126 PISA
Italy
phone: +39-50-593435
fax: +39-50-554342
e.mail: tonazzini@iei.pi.cnr.it
url: http://www.iei.pi.cnr.it/People/Tonazzini.html